Science and Engineering at The University of Edinburgh

Biomass estimation of tropical wooded savanna using radar remote sensing. - (PhD project started September 2004, planned completion October 2007)
This project aims to investigate the use of radar remote sensing techniques for estimating the biomass of wooded savanna. The objective is to investigate relationships between radar backscatter and several physical characteristics of tree and palm groupings in a wooded savanna. The test site is a well-studied tract of tropical savanna in the Rio Bravo Conservation & Management Area in northwest Belize. Field data will be collected during April and May 2005, which coincides with the dry season of the study area. Radar data is already available (see data pages).

Tree height as an indicator of biomass - An INSAR approach. - (MSc project, started January 2005, planned completion September 2005)
This study will assess the use of SAR Interferometry (INSAR) for estimating tree height. Through the use of established allometric relationships, the tree height estimates will be converted to biomass estimates. The project will make use of airborne X-band and C-band Interferometric data to provide height estimates; field validation will be carried out during April 2005.

The project will take advantage of two sets of high-quality airborne Synthetic Aperture Radar (SAR) data recently acquired over the test site in Belize to answer the following research questions:

• At what level of biomass density (if at all) does the backscatter response in L and P band saturate upon interaction with the typical components of savanna vegetation in Belize?
• How do estimates of the height of savanna vegetation canopies obtained from X- and C- band compare to radar intensity (at L- and P-band) for estimating above-ground biomass?
• Can these results be extrapolated to predict the ability of spaceborne radar systems (both current and future) to identify the components of savanna land cover and estimate their carbon storage?

Radar data available over the study area include a JPL-Airsar "opportunity" acquisition with C-band single-pass interferometry, and L- and P-band polarimetry (March 2004) and X-band airborne SAR data acquired by Intermap Technologies using STAR3i (April 1999). This first commercial SAR data for Belize includes a first return Digital Surface Model (DSM) from the "near-top" of the vegetation canopy (5m grid, ~3m RMSE vertical) and an orthorectified image (ORI) of backscatter intensity (2.5m). Several archived Landsat images and an earlier Airsar mission from 1994 are also available.

The knowledge and experience gained from conducting this pilot-study in Belize will provide a more practical understanding of the potential of spaceborne radar for mapping and monitoring savanna ecosystems worldwide. In so doing, this project will support NERC priority issues of developing methodologies for monitoring global change and improving exploitation of current and future spaceborne observations (Earth Observation Strategy). It will also provide the first comprehensive and detailed vegetation inventory and terrain model for the savannas of Belize based on interferometric SAR, which will directly benefit the local NGOs responsible for savanna management in Belize.

Improving savannah vegetation classification using topographic data. - 2004.

Building on previous work using medium resolution optical Landsat imagery this project investigated the use of radar backscatter and surface elevation data to investigate internal savanna composition. Accurate delimitation of savanna formations is of importance to land managers, and often visual image interpretation is used to aid savanna management. However the heterogeneity of these formations and frequent cloud cover limits the use of such data in this context. SAR imagery by contrast operates day or night, can penetrate through clouds and planned satellites will be capable of supplying data at spatial resolutions similar to available airborne SAR scenes.

A combination of SAR backscatter intensity and surface elevation data at 2.5m posting was visually assessed to identify the extent of pine/oak formations. It was found that these formations could be clearly and accurately delimited within the study area. Future work will examine the extensibility of a visual savanna classification methodology and its usability for land managers.

Synthetic Aperture Radar for neo-tropical vegetation height retrieval. - 2004.
Following on from work in the UK, the aim of this pilot-study research was to assess the capability of airborne short wavelength Synthetic Aperture Radar (SAR) for estimating neo-tropical savannah vegetation height. The objectives to reach this aim were to:

• Conduct fieldwork to measure accurate topographic data in a series of profiles and spot heights.
• Collect an inventory of vegetation types along these transects.
• Collect height information from the different vegetation types, with an emphasis on high forest stands.
• Validate the accuracy of the SAR derived Digital Elevation Model (DEM) against true elevation.
• Compare SAR estimated vegetation height against true height.

Results